Description of problem: ------------------------ The fix for the bug 2091976 was not backported to CNV v4.12.z. This is required for CNV v4.12.z Version-Release number of selected component (if applicable): ------------------------------------------------------------- CNV 4.12.3 How reproducible: ----------------- Always Steps to Reproduce: ------------------- 1. Create a VM with memory.request as 1Gi and memory.limits as 2Gi 2. Run stress-ng inside the VM, to make use of 100% of memory Actual results: --------------- Alert is thrown only when free memory goes below 20MB, which is very close to the VM getting evicted. It is because of the reason this limit is raised to 50MB Expected results: ----------------- Alert 'KubevirtVmHighMemoryUsage' should be triggered when the free memory of the VM goes below 50MB. The patch that addressed the issue for the bug 2091976 should also be backported to CNV v4.12
Verified with CNV v-4.12.4 interim version ( 4.12.4-50 ) with the following steps. 1. Created a fedora VM with memory requests set to 1Gi and memory limits set to 2Gi 2. Identify the virt-launcher pod. 3. Decrease the free/available memory in that virt-launcher pod. # cat <( </dev/zero head -c 943718400) <(sleep 150) | tail 4. Check for alert in OpenShift Web console. Following alert message was seen: "Container compute in pod virt-launcher-fedora-044iv4ww85x3w21a-kh4tn in namespace testproject free memory is less than 20 MB and it is close to requested memory" But to clarify this the check was actually made for 50MB, though the alert message describes it as 50MB. This is because of the missing patch.Bug 2210988 takes care of updating this value to 50MB in all the versions. With this information marking this bug as verified.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (OpenShift Virtualization 4.12.4 Images), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHEA-2023:3889